AI Is Producing Motion. The Return Depends on Meaning.

The Wall Street Journal's chief economics commentator, Greg Ip, recently asked a question every leader should sit with: Will America spend 9% of its GDP on AI? Drawing on research by Columbia finance professor Stijn Van Nieuwerburgh, he reports that to justify the sums now being committed to AI, Americans would eventually need to spend roughly $3.5 trillion a year on AI services by 2032. That is about as much of our income as we spend on food.

It is a careful, balanced piece. Ip lays out the skeptic's case: diminishing returns, and prices that tend to collapse as capacity grows. He lays out the optimist's case too: falling prices and rising capability can create more demand than they lose. And he ends on a sentence I have not stopped thinking about. AI will almost certainly raise productivity, he writes, but not necessarily in ways that translate directly into revenue.

That sentence is where the real story begins.

The Study Hiding in Plain Sight

Among the evidence Ip cites is a study by Harvard researchers Fiona Chen and James Stratton, who examined how engineering teams actually work once AI arrives. Their findings, published in Artificial Intelligence in the Firm: Bottlenecks in Software Production, are remarkable. AI agents increased lines of code by 30% and pull requests by 23%.

And yet completed projects barely moved. The effect was small and statistically insignificant.

Why? Because the work did not disappear. It moved. Review time rose 49%, from about seven days to more than ten. The number of submissions needing revisions nearly doubled. The machines produced more. The people around them had to judge, correct and integrate more, and the whole system slowed at the point where human judgment was required.

This is what the headline misses. The investment math prices the machine. The return depends on the human system around it.

Motion Is Not Meaning

I have spent two decades inside organizations watching a pattern repeat. A new tool arrives, activity rises, dashboards light up, and leaders celebrate progress. But a well-designed system can give the illusion of progress. More output is not the same as more value.

AI makes that illusion easier to believe than ever, because the motion is real and measurable. Lines of code, documents drafted, reports generated, tickets closed. What is harder to measure is meaning: whether the work is right, whether it fits the moment, whether anyone has the conviction to say it is not good enough yet.

Meaning is where the bottleneck lives. And meaning is human work.

That is why the most important AI decisions are not about models or data centers. They are about whether leaders redesign the work around the machine, and whether they protect the judgment the machine cannot supply. As I explored in AI Brain Fry Is Really a Judgment Problem, the deeper risk is not that AI does too little. It is that people stop thinking before the tool does.

From Results to Methods

In my work on Leadership in the Age of Personalization, one of the five shifts every organization must make is from results to methods. The question beneath it is simple: How do you let them do it?

Most organizations are measuring AI by results, the volume of output it produces. The Harvard study suggests the value is decided by methods: how review works, who is trusted to make the call, how people collaborate with the tool and with each other. Standardized metrics still matter. They simply need to be modernized for a world where the scarce resource is no longer production but discernment.

There is evidence that this choice shows up in revenue. An Orgvue analysis of Fortune 500 companies reported in July 2026 found that companies investing in human-fueled growth delivered nearly twice the revenue growth of companies trying to do more with less. The organizations betting only on the machine may be leaving the largest return on the table. I made a similar case in You Are What the AI Efficiency Calculation Leaves Out.

Three Questions for Leaders This Week

Whether AI ultimately reaches 9% of GDP is a question for economists and markets. Whether it creates value inside your organization is a question for you. Start with three questions.

Where is AI producing more motion than meaning? Look for places where activity has risen but outcomes have not.

Where has the bottleneck moved to human judgment? Those are the people and processes that now deserve your investment, not your neglect.

Are we measuring output, or how the work actually gets done? If your dashboards only count volume, they will keep telling you a story of progress that the results may not confirm.

The AI boom will not be justified by what the machines can do. It will be justified by what leaders choose to do next. Transformation does not fail from lack of vision. It fails from lack of daily practice, and the practice that matters most now is human.

Want to explore these ideas further? Learn more about my work on leadership, identity and conviction at www.theglennllopis.com.